Serverless Agentic Workflows with Amazon Bedrock
by DeepLearning.AI x AWS
Build one customer-service agent lesson by lesson until it runs serverless, with guardrails and a knowledge base attached.
Overview
Taught by Mike Chambers, a senior developer advocate for generative AI at AWS, this DeepLearning.AI short course is built around one continuous build: a customer-service agent for a fictional tea-mug retailer that you extend lesson by lesson until it runs as a deployable serverless application. The eight lessons total about one hour and thirty-four minutes, and six of them ship runnable notebooks. You start by creating an agent with Amazon Bedrock Agents and giving it an action group, then connect it to a mock CRM so it can look up real customer records, add the Bedrock code interpreter so the agent performs arithmetic in Python instead of hallucinating totals, apply Bedrock Guardrails to block malicious prompts and shield customers from harmful or sensitive content, and finally attach a Bedrock knowledge base so the agent answers policy questions from an FAQ document rather than from the model weights. A console walkthrough then rebuilds the same agent through the AWS UI, which matters because the SDK and the console do not look alike. The course is deliberately operational rather than conceptual: the subject is the scaffolding an agent needs to survive contact with users - tool wiring, safety filters, retrieval, and infrastructure that scales to zero - not agent theory. It is free to take on the DeepLearning.AI platform and mirrored as a Coursera guided project; a completion certificate requires a paid PRO membership.
At a Glance
- Topic
- Agentic
- Level
- Intermediate
- Format
- Course
- Cost
- Free
- Duration
- ~1 hour 34 minutes, self-paced (8 lessons, 6 with runnable code)
- Provider
- DeepLearning.AI x AWS
- Hands-on
- Yes — code/exercises
- Certificate
- None
What You’ll Learn
- ✓Create an Amazon Bedrock agent and attach action groups so it can call tools
- ✓Wire an agent to a CRM so it retrieves real customer records at request time
- ✓Enable the Bedrock code interpreter so the agent computes instead of guessing numbers
- ✓Apply Bedrock Guardrails to block malicious prompts and shield sensitive customer data
- ✓Attach a Bedrock knowledge base so answers are grounded in source documents
- ✓Deploy the finished agent on serverless infrastructure that scales down to zero
- ✓Rebuild the same agent through the AWS console rather than the notebook SDK
Highlights
- •One continuous build rather than six disconnected demos - the same support agent gains a capability each lesson
- •Guardrails and sensitive-data masking get a full lesson, not a closing caveat
- •Covers both the notebook SDK path and the AWS console path, which diverge in practice
- •Under 100 minutes end to end and free to take, so the time cost of trying it is near zero
- •Taught by an AWS developer advocate, so the Bedrock service details are first-hand
Who It’s For
Best For
- ✓AWS engineers who need agents running in production, not in a notebook
- ✓Backend developers evaluating Bedrock Agents against LangChain, CrewAI or LangGraph
- ✓Teams that must show a safety and PII story before an agent ships to customers
Prerequisites
- •Intermediate Python - you read and modify notebooks rather than write from scratch
- •An AWS account and basic familiarity with the console, IAM and serverless concepts
FAQ
What is Serverless Agentic Workflows with Amazon Bedrock?
A short, hands-on DeepLearning.AI course built with AWS that walks you through creating an Amazon Bedrock agent, connecting it to a CRM and a code interpreter, wrapping it in guardrails that mask sensitive data, and grounding it in a knowledge base. By the end you have deployed a responsible agentic application on serverless infrastructure that scales to zero, and you know which pieces belong around an agent in production rather than in a notebook.
Is Serverless Agentic Workflows with Amazon Bedrock free?
Serverless Agentic Workflows with Amazon Bedrock is free to access.
What level is Serverless Agentic Workflows with Amazon Bedrock for?
Serverless Agentic Workflows with Amazon Bedrock is aimed at a intermediate audience. Recommended background: Intermediate Python - you read and modify notebooks rather than write from scratch, An AWS account and basic familiarity with the console, IAM and serverless concepts.
How long does Serverless Agentic Workflows with Amazon Bedrock take?
Expect roughly ~1 hour 34 minutes, self-paced (8 lessons, 6 with runnable code). Most learners work through it at their own pace.
What will I learn from Serverless Agentic Workflows with Amazon Bedrock?
You'll learn: Create an Amazon Bedrock agent and attach action groups so it can call tools; Wire an agent to a CRM so it retrieves real customer records at request time; Enable the Bedrock code interpreter so the agent computes instead of guessing numbers; Apply Bedrock Guardrails to block malicious prompts and shield sensitive customer data; Attach a Bedrock knowledge base so answers are grounded in source documents; Deploy the finished agent on serverless infrastructure that scales down to zero; Rebuild the same agent through the AWS console rather than the notebook SDK.
Topics
Sources
This page was written from 3 sources, 2 on domains other than deeplearning.ai.